The EGFR-MAPK pathway is a key signaling pathway in human Triple Negative Beast Cancers (TNBC). We propose to leverage genomic and proteomic data from a rich animal model system, and 2 human clinical trials, to build predictive models of the EGFR-MAPK signaling pathway activity for TNBC patients. The heterogeneity of TNBC has hindered previous development of predictive pathway-based computational models because most approaches are based on experimental data from a single cell line or animal model that is then extrapolated to fit multiple tumor subtypes. Our approach is to use a diverse experimental model system that reflects the heterogeneous disease subtypes, and then use two distinct and complementary methods to build the computational model. We will simultaneously use mechanistic and statistical modeling approaches, at a variety of scales, that incorporate data from drug treated tumors and cell lines, assayed for gene expression, DNA copy number, DNA mutations, and protein kinome activity. Lastly, we will test these computational models on human tumors to evaluate their predictive performance.